Analysis of the laser system design was carried out leading to a system model and a diagnostic matrix mapping possible faults to observable symptoms. Two approaches to symptom monitoring were evaluated. Firstly, a classical statistical technique (the Control Chart or Shewhart) was combined with a rule-based system, and implemented in custom software that could be embedded in the product. This enabled continuous condition monitoring of the system. Secondly, an artificial intelligence technique known as a 'neural network' was evaluated as a possible method of monitoring the output of the laser in such a way that potential failure could be anticipated. The techniques made available to
the Company through the project satisfied the Target Requirement. Embedding of the knowledge was achieved through the Associate carrying out a programme of training of company staff in the new techniques and updating company documentation.